Hi All, I have a production firewall on a Sun V120 running OpenBSD 4.5 sparc64, with 2 active interfaces. Two weeks ago, the gem1 interface suddenly hung and I was able to revive it using "ifconfig gem1 down; ifconfig gem1 up". I found the following m...@openbsd thread from March 2009:
http://www.mail-archive.com/[email protected]/msg73257.html After contacting Daniel, it seems the issue was fixed in a newer gem.c. But since this appeared to be specific to the gem/eri hardware, I installed a quad hme card and moved over to hme0 and hme1. Today the same thing happened with hme1, and the ifconfig down/up brought it back. System was only up for 1 day when this happened. Where should I be looking to track down this problem? With gem I noticed some Ierrs in netstat but no Oerrs. With hme, I see Oerrs but no Ierrs. Either way, on the switch (Cisco 2924) I get a handful of input errors. The /var/log/messages shows nothing. I thought maybe it's a cable or switch problem, but then why would the ifconfig down/up in OpenBSD bring it back without touching the cable or switch at all? I swapped the hme1 cable today just to see if it has any effect on the error counters. netstat -i after each failure: gem1 1500 <Link> 00:03:ba:ce:da:8a 1463892401 14 1119483645 0 0 hme1 1500 <Link> 08:00:20:ee:8d:4d 31718821 0 31632599 28 0 switch interface status after hme failure: 23870059 packets input, 63580130 bytes Received 124 broadcasts, 0 runts, 0 giants, 0 throttles 26 input errors, 26 CRC, 0 frame, 26 overrun, 26 ignored 0 watchdog, 0 multicast 0 input packets with dribble condition detected 24011555 packets output, 2096322509 bytes, 0 underruns 0 output errors, 0 collisions, 2 interface resets 0 babbles, 0 late collision, 0 deferred 0 lost carrier, 0 no carrier 0 output buffer failures, 0 output buffers swapped out Bryan
